Dietary carbohydrates, fiber, and breast cancer risk in Chinese

women13
Wanqing Wen, Xiao Ou Shu, Honglan Li, Gong Yang, Bu-Tian Ji, Hui Cai, Yu-Tang Gao, and Wei Zheng
ABSTRACT
Background: Few studies have investigated the association of dietary carbohydrate and fiber intake with breast cancer risk in women
in China, where carbohydrate intake is traditionally high.
Objective: The objective was to prospectively evaluate the association of dietary carbohydrates, glycemic index, glycemic load, and
dietary fiber with breast cancer risk and to determine whether the effect
of these dietary intakes is modified by age and selected insulin- or
estrogen-related risk factors.
Design: A total of 74,942 women aged 4070 y were recruited into
the Shanghai Womens Health Study, a population-based cohort study.
Dietary intake was assessed by in-person interviews. A Cox proportional hazards regression model was used to evaluate associations.
Results: During an average of 7.35 y of follow-up, 616 incident breast
cancer cases were documented. A higher carbohydrate intake was
associated with a higher risk of premenopausal breast cancer (P for
trend 0.002). Compared with the lowest quintile, the hazard ratios
(and 95% CIs) were 1.47 (1.00, 2.32) and 2.01 (1.26, 3.19) for the
fourth and fifth quintiles, respectively. A similar pattern was found for
glycemic load. The association between carbohydrate intake and
breast cancer was significantly modified by age; the increased breast
cancer risk associated with carbohydrate intake was restricted to
women who were younger than 50 y. No significant association of
breast cancer risk with glycemic index or dietary fiber intake was
found.
Conclusion: Our data suggest that a high carbohydrate intake and
a diet with a high glycemic load may be associated with breast
cancer risk in premenopausal women or women ,50 y.
Am J
Clin Nutr 2009;89:2839.

INTRODUCTION

Recently, the association of dietary carbohydrate intake with

breast cancer risk has received a significant amount of attention
based on the hypothesis that a high dietary carbohydrate intake
increases insulin resistance and increases plasma concentrations of
insulinconditions that may be associated with breast cancer risk
(14). The specific composition of dietary carbohydrates consumed may be particularly important in breast cancer risk. Consumption of high-glycemic-index foods was associated with
higher postprandial or fasting plasma insulin concentrations (5, 6).
We recently reported that a high intake of foods with a high glycemic index and a high glycemic load increases the risk of type 2
diabetes mellitus in Chinese women (7). Some case-control
studies (810) and a prospective study (11) have reported that

high-glycemic-index foods may be associated with an increased

risk of breast cancer. Some prospective studies have shown that the
effect of carbohydrates on breast cancer risk may be modified by
nondietary risk factors that are associated with estrogen metabolism and insulin resistance, such as menopausal status, body
weight, and physical activity (1113) Most prospective studies,
however, have shown no overall association between dietary
carbohydrate intake and breast cancer risk (1217).
Dietary fiber, on the other hand, may decrease breast cancer risk
through mechanisms that decrease circulating estrogens (18, 19).
A considerable part of the Chinese diet consists of resistant starch,
which contributes to dietary fiber and has a low glycemic index.
Epidemiologic studies have reported inconsistent findings on
these hypotheses. Some case-control studies have provided evidence to support these hypotheses (20, 21). Some recent prospective studies have reported inverse associations between
dietary fiber intake and breast cancer risk (22, 23), whereas others
have reported no association (12, 14, 24).
Traditional ecological analysis suggests a positive correlation
between national per capita fat consumption and breast cancer
incidence and mortality (25). Chinese women historically have
consumed a high amount of carbohydrates, yet have a low risk of
breast cancer. As rapid economic development continues in
Shanghai, China, there have been substantial dietary changes
among Chinese women characterized by increasing fat intake and
decreasing carbohydrate intake (26, 27). Coincidently, Chinese
women in Shanghai have experienced a dramatic increase in
breast cancer incidence in recent years (27). However, these ecological correlations were generally not confirmed by epidemiologic studies, most of which were conducted in Western
populations, where fat intake is high but carbohydrate intake is
low. This study investigated the association of dietary carbohydrate with breast cancer risk in a Chinese population in whom
1
From the Vanderbilt Epidemiology Center, Vanderbilt University Medical Center, Nashville, TN (WW, XOS, GY, HC, and WZ); the Department of
Epidemiology, Shanghai Cancer Institute, Peoples Republic of China (HL
and Y-TG); and the Division of Cancer Epidemiology and Genetics, National
Cancer Institute, NIH, DHHS, Bethesda, MD (B-TJ).
2
Supported by US Public Health Service grant number R01 CA070867.
3
Address reprint requests and correspondence to W Wen,Vanderbilt Epidemiology Center, Institute of Medicine & Public Health, Vanderbilt University Medical Center, Sixth Floor, Suite 600, 2525 West End Avenue,
Nashville, TN 37203-1738. E-mail: wanqing.wen@vanderbilt.edu.
Received May 1, 2008. Accepted for publication October 19, 2008.
First published online December 3, 2008; doi: 10.3945/ajcn.2008.26356.

Am J Clin Nutr 2009;89:2839. Printed in USA. 2009 American Society for Nutrition

283

284

WEN ET AL

carbohydrate intake is traditionally high but which is decreasing in

younger women (26) and may provide an opportunity to disprove
the traditional ecological analysis of negative correlation between
carbohydrate intake and breast cancer.
In the present study, we prospectively evaluated the association
of dietary carbohydrate intake, glycemic index, glycemic load, and
dietary fiber intakewith breast cancer risk inthe Shanghai Womens
Cohort Study (SWHS). In addition, we investigated whether the
effect of these dietary intakes are modified by age and selected
insulin- or estrogen-related risk factors in women.

(32). Glycemic index values for individual food items from published data were added to the nutrient database (7). A foods
glycemic load was calculated by multiplying the carbohydrate
content of each food by its glycemic index values and its frequency of consumption (33). Dietary glycemic load for a participant was calculated by summing the values of glycemic load for
all food items. Overall dietary glycemic index for a participant
was calculated by dividing glycemic load by the amount of carbohydrates consumed.
Identification of breast cancer cases

SUBJECTS AND METHODS

Study participants
The SWHS, initiated in March 1997, is an ongoing prospective
cohort study being conducted in urban Shanghai, China. The study
was approved by the relevant institutional review boards for human
research, and the details of the study design were described
elsewhere (28, 29). Briefly, 81,170 women aged 4070 y who
resided in 7 geographically defined communities in urban Shanghai
were approached, and 75,221 participated in the study; the response rate was 92.7%. Of those who completed the survey, 279
women were later found to be younger than 40 y or older than 70 y
at the time of the baseline interview and thus were excluded from
the cohort. The remaining 74,942 women constituted the cohort.
All subjects were interviewed in person by trained interviewers
using a questionnaire. The questionnaire included, among other
items, questions on sociodemographic factors, dietary and lifestyle
habits, menstrual and reproductive history, hormone use, and
medical history. Anthropometric measurements, including current
weight, height, and circumferences of the waist and hips, were
made by trained interviewers who are retired health professional
according to standard procedures (30). All measurements were
taken twice with a preset tolerance of 1 kg for weight and 1 cm for
height, waist, and hip circumferences. A third measurement was
taken if the difference between the 2 measurements was larger than
the tolerance limit. The averages of the 2 closest measurements
were used in the current analysis.
Dietary assessment
A quantitative food-frequency questionnaire (FFQ) was used to
assess usual dietary intake at the baseline survey and again at the
first follow-up survey conducted 23 y after the baseline measurement. The FFQ was validated against the average of multiple
24-h dietary recalls. The correlation coefficients for macronutrients between the FFQ and the 24-h recall ranged from 0.59 to
0.66 (28). The FFQ covered .90% of foods commonly consumed
in urban Shanghai (28, 29). During the in-person interviews, each
participant was first asked how often, on average, during the past
12 mo she had consumed a specific food or food group (the
possible responses ware daily, weekly, monthly, yearly, or never)
and then how much they consumed in grams per unit of time.
Macronutrient intakes from each food were calculated by multiplying the amount of food consumed by the nutrient content per
gram of the food, as obtained from the Chinese food-composition
tables (31). The total dietary intake of macronutrients was calculated by summing across all food items.
The glycemic index ranks foods on the basis of the relative
postprandial blood glucose response per gram of carbohydrate

The cohort has been followed by a combination of in-person

surveys and periodic linkage with records kept by the Shanghai
Tumor Registry. Every 2 y, the cohort members are interviewed to
record details of their interim health history, including their history
of cancer, cardiovascular disease, stroke, and other chronic diseases. Every year, we conduct a record linkage of cohort member
information with the cancer registry and death certificate registry to
ensure a timely and complete ascertainment of new cancer cases
and deaths in the study cohort. All possible matches are checked
manually and are verified through home visits. Copies of medical
charts are obtained to verify cancer diagnoses and to collect detailed information on the pathological characteristics of cancers.
Statistical analysis
The end date of the observation in this study was set as the date of
cancer diagnosis or the date of death for deceased cohort members
or the date of the last follow-up. For those participants whose last inperson contact was before 31 December 2005, the end date of the
observation was set as 31 December 2005, 1 y before the most
recent record linkage, to allow for delays in records processing.
Hazard ratios and 95% CIs were estimated by using the Cox
proportional hazards regression model (34). Survival was modeled as a function of age and stratified by birth cohort intervals
to control for age and cohort effects (35). Left truncation was set
to the date of the baseline survey.
Dietary intake estimated from only the baseline interview and
the average dietary intake calculated from both the baseline interview and the first follow-up interview were analyzed. To prevent
bias due to dietary changes related to disease development, the
baseline dietary intake was used to replace the average dietary
intake for study participants who were diagnosed with cardiovascular diseases, cancers, and diabetes within 1 y before the first
follow-up interview.
Carbohydrate and fiber intakes, glycemic index, and glycemic
load were adjusted for total energy intake with the regressionresidual method (36). The energy-adjusted dietary intakes were
categorized into quintiles and also analyzed as continuous variables to evaluate linear trends. The interaction of continuous dietary intake variables with age under observation and other
common risk factors for breast cancer risk was evaluated with
a likelihood ratio test by comparing the Cox models with and
without the interaction term. Information on menopausal status
was updated at each follow-up interview. Menopausal status and
age of observation were treated as time-varying variables when
evaluating their interaction with dietary intakes (34). In addition to
total energy intake, age at start of follow-up, education level, body
mass index (BMI), age at first birth, breast cancer history in

285

CARBOHYDRATES, FIBER, AND BREAST CANCER RISK

are shown in Table 1. Consistent with previous study findings,

women who were obese, were older at first birth, had higher levels
of education, had a history of breast cancer in a first-degree relative, had a personal history of benign breast disease, or were
physically inactive had an increased risk of breast cancer. These
risk factors were adjusted for with a Cox regression model in the
analyses of dietary intakes.
The association of breast cancer risk with carbohydrate intake,
glycemic index, glycemic load, and fiber intake measured at the
baseline interview by quintiles for all women in the cohort and
separated by menopausal status are shown in Table 2. Carbohydrate intake was significantly associated with elevated breast
cancer risk in premenopausal women (P for linear trend 0.001).
Compared with the lowest quintile, the hazard ratios (and 95%
CIs) were 1.47 (1.00, 2.32) for the fourth quintile and 2.01 (1.26,
3.19) for the fifth quintile. However, this association was not seen
in postmenopausal women. A similar association pattern was
found for glycemic load, which was highly correlated with

first-degree relatives, personal history of benign breast diseases,

and physical activity were also adjusted for in the analyses.
RESULTS

For the current analysis, we excluded 1576 women with

a history of cancer at baseline, 28 women with outlier values for
dietary intake (.3 SDs from the mean; ie, women with a carbohydrate intake ,50 or .705 g/d or a total energy intake ,388
or .4577 kcal/d), and 10 women who were lost to follow-up
shortly after study enrollment. We documented 616 incident
breast cancer cases in the remaining 73,328 women during an
average of 7.35 y of follow-up (538,899 person-years). Of 616
incident breast cancer cases, 593 cases were invasive carcinoma,
23 cases were in situ carcinoma, 190 cases occurred before
menopause, and 426 cases occurred after menopause.
The characteristics of women in the SWHS cohort and the association of known nondietary risk factors with breast cancer risk

TABLE 1
Baseline characteristics of women in the Shanghai Womens Cohort Study and the association of breast cancer risk with
common risk factors
Variables

Value
(n 73,328)
2

Age (y)
52.5 6 9.1
49.1 6 4.0
Age at menopause (y)3
Energy intake (kcal/d)
1675.4 6 400.5
Carbohydrate intake
68.5 6 6.9
(% of energy)
Energy-adjusted glycemic index
70.7 6 5.2
Energy-adjusted glycemic load
202.0 6 31.2
Fiber intake (g/100 kcal)
11.0 6 3.3
Waist-to-hip ratio
0.81 6 0.05
.0.81

BMI (kg/m2)
24.0 6 3.4
.25

Age at menarche
14.9 6 1.7
.15 y

1415 y

13 y

Age at first birth

25.6 6 4.1
,25 y

2529 y

30 y

Nulliparous

Education level
,Middle school

Middle school

High school

.High school

Breast cancer history in first-degree relatives

Personal history of benign breast disease

Physical activity

Hormone replacement therapy use

Ever smoked

Ever drank alcohol

Percentage of study
population (n 73,328)

HR
(95% CI)1

P for trend

47.9

35.1

35.9
42.1
22.1

39.2
46.0
11.5
3.3

1.20 (1.02, 1.41)

1.31 (1.11, 1.55)

1.00 (reference)
1.07 (0.89, 1.30)
1.08 (0.87, 1.35)

1.00 (reference)
1.22 (0.99, 1.50)
1.56 (1.19,2.03)
1.46 (0.96, 2.23)

0.347

0.0024

21.4
37.2
27.9
13.5
1.9
16.9
35.1
2.1
2.8
2.3

1.00 (reference)
1.52 (1.13, 2.07)
2.21 (1.63, 3.00)
1.97 (1.41, 2.75)
2.14 (1.45, 3.15)
1.60 (1.32, 1.94)
0.85 (0.71, 1.00)
1.10 (0.68, 1.77)
1.25 (0.68, 2.28)
1.15 (0.63, 2.10)

,0.001

The hazard ratios (HRs) and 95% CIs were derived from the Cox regression model with age as the time scale. All
nondietary risk factors listed were included in the same model for mutual adjustment.
2
Mean 6 SD (all such values).
3
Among postmenopausal women.
4
For parous women only.

286

WEN ET AL
TABLE 2
Hazard ratios (and 95% CIs) for the association of dietary carbohydrate and fiber intake with breast cancer risk by
quintile (Q)1

Carbohydrate intake
Q1
Q2
Q3
Q4
Q5
P for trend
P for interaction
Glycemic index
Q1
Q2
Q3
Q4
Q5
P for trend
P for interaction
Glycemic load
Q1
Q2
Q3
Q4
Q5
P for trend
P for interaction
Fiber intake
Q1
Q2
Q3
Q4
Q5
P for trend
P for interaction

Premenopausal women
(n 190)2

Postmenopausal women
(n 426)2

Median value

All subjects
(n 616)2

257.5
263.2
273.8
289.3
343.5

1.00 (reference)
1.06 (0.82, 1.37)
1.06 0.82, 1.36)
1.08 (0.83, 1.39)
1.22 (0.94, 1.58)
0.204

1.00
1.17
1.11
1.47
2.01

(reference)
(0.74, 1.85)
(0.69, 1.77)
(1.00, 2.32)
(1.26, 3.19)
0.001

1.00
1.02
1.03
0.93
0.98

(reference)
(0.75, 1.38)
(0.76, 1.39)
(0.68, 1.27)
(0.72, 1.34)
0.549
0.001

63.9
68.5
71.2
73.6
76.8

1.00
1.09
1.01
0.93
1.03

(reference)
(0.85, 1.38)
(0.79, 1.29)
(0.72, 1.20)
(0.79, 1.34)
0.472

1.00
0.97
1.08
1.39
1.19

(reference)
(0.62, 1.51)
(0.70, 1.68)
(0.90, 2.13)
(0.73, 1.94)
0.256

1.00
1.14
0.98
0.76
0.96

(reference)
(0.85, 1.52)
(0.72, 1.32)
(0.55, 1.05)
(0.70, 1.31)
0.093
0.068

163.8
187.5
202.5
216.7
239.4

1.00
1.02
0.99
1.07
1.07

(reference)
(0.79, 1.30)
(0.77, 1.27)
(0.83, 1.38)
(0.82, 1.39)
0.552

1.00
0.86
0.93
1.63
1.53

(reference)
(0.54, 1.37)
(0.59, 1.48)
(1.07, 2.48)
(0.96, 2.45)
0.008

1.00
1.08
1.00
0.86
0.91

(reference)
(0.80, 1.45)
(0.74, 1.35)
(0.63, 1.18)
(0.67, 1.25)
0.291
0.007

7.7
8.8
10.2
12.0
16.3

1.00
1.05
1.03
1.01
1.09

(reference)
(0.81, 1.35)
(0.80, 1.33)
(0.78, 1.31)
(0.84, 1.40)
0.482

1.00
0.95
0.85
0.71
1.01

(reference)
(0.61, 1.49)
(0.54, 1.34)
(0.44, 1.15)
(0.64, 1.57)
0.635

1.00
1.08
1.13
1.17
1.12

(reference)
(0.79, 1.49)
(0.82, 1.54)
(0.86, 1.59)
(0.83, 1.53)
0.267
0.225

1
The hazard ratios and 95% CIs were derived from the Cox regression model with age as the time scale and were
adjusted for age at the start of follow-up, total energy intake, education level, BMI, age at first birth, breast cancer history in
first-degree relative, personal history of benign breast diseases, and physical activity.
2
Number of breast cancer cases.

carbohydrate intake in this study (correlation coefficient 0.95).

No significant association of breast cancer risk with glycemic
index or fiber intake was found in either premenopausal women or
postmenopausal women.
We examined whether the association between carbohydrate
intake and breast cancer risk was modified by age or other riskfactors
for breast cancer, including education levels, personal history of
benign breast disease, and selected insulin- or estrogen-related risk
factors, including BMI,physical activity, age atmenarche, and age at
first birth. As presented in Table 3, the association of dietary carbohydrate intake with breast cancer was significantly modified by
age (P 0.002 in all women, P 0.012 in premenopausal women).
The increased breast cancer risk associated with carbohydrate intake was restricted to women who were ,50 y. No age-carbohydrate interaction was found in postmenopausal women (P 0.940;
data not shown). We found no significant interaction between carbohydrate intake and other factors when the analyses included all
women, premenopausal women only (as presented in Table 3), or
postmenopausal women only (data not shown).
We conducted similar analyses for the average dietary intake
calculated from both the baseline interview and the first follow-up

interview. We also conducted analyses restricted to invasive breast

cancer cases only and excluding breast cancer cases that occurred
within 1 y of the baseline interview to minimize the possible effect
of dietary changes related to subclinical breast cancer. All of these
analyses produced results similar to those reported here.
DISCUSSION

In this prospective study we found that a high dietary carbohydrate intake and glycemic load were associated with elevated
breast cancer risk in premenopausal, but not in postmenopausal,
women in a dose-response manner. We found no clear association
between glycemic index or dietary fiber intake and breast cancer
risk.
Using an energy-partition model (36) to examine the effect of
dietary fat, protein, and carbohydrate intakes, we found that the
effect of carbohydrate intake was opposite that of dietary fat and
protein intakes. Using a residual model to adjust for total energy
intake, we interpreted the coefficients for carbohydrate intake as
the effect of substituting a certain amount of carbohydrate (eg, 50 g)
for the same amount of energy from noncarbohydrate sources (ie,

287

CARBOHYDRATES, FIBER, AND BREAST CANCER RISK

TABLE 3
Association of dietary carbohydrate intake with breast cancer risk by common risk factors
All women

Overall
Age
,50 y
5059 y
60 y
Education level
High school
.High school
BMI (in kg/m2)
25
.25
Physical activity
No
Yes
Personal history of benign
breast diseases
No
Yes
Age at menarche
.15 y
1415 y
13 y
Age at first birth
,25 y
2529 y
30 y or nulliparous

Premenopausal women

No. of breast
cancer cases

HR (95% CI)1

P2

No. of breast
cancer cases

HR (95% CI)1

P2

616

1.07 (0.93, 1.24)

190

1.59 (1.20, 2.10)

179
207
230

1.72 (1.28, 2.31)

0.83 (0.66, 1.05)
1.01 (0.80, 1.27)

0.002

137
53
0

1.84 (1.32, 2.59)

1.11 (0.67, 1.84)

0.012

282
334

1.23 (0.99, 1.51)

0.97 (0.80, 1.19)

0.295

86
104

1.90 (1.24, 2.91)

1.41 (0.97, 2.04)

0.638

377
239

1.09 (0.90, 1.31)

1.06 (0.85, 1.31)

0.590

143
47

1.54 (1.10, 2.16)

1.71 (1.05, 2.80)

0.334

415
201

1.11 (0.93, 1.33)

0.99 (0.78, 1.27)

0.788

153
37

1.57 (1.15, 2.14)

1.70 (0.89, 3.26)

0.222

461
155

1.14 (0.97, 1.35)

0.89 (0.66, 1.18)

0.506

129
61

1.76 (1.26, 2.47)

1.25 (0.75, 2.07)

0.369

191
268
157

1.09 (0.85, 1.39)

1.15 (0.92, 1.43)
0.96 (0.72, 1.28)

0.151

51
93
46

1.91 (1.13, 3.25)

1.44 (0.97, 2.15)
1.51 (0.83, 2.74)

0.237

199
294
123

1.22 (0.95, 1.57)

1.10 (0.89, 1.36)
0.87 (0.64, 1.17)

0.750

23
120
47

2.30 (0.98, 5.39)

1.80 (1.26, 2.60)
1.12 (0.68, 1.85)

0.714

The hazard ratios (HRs) and 95% CIs were derived from the Cox regression model with age as the time scale and were
adjusted for age at start of follow-up, total energy intake, education level, BMI, age at first birth, breast cancer history in firstdegree relative, personal history of benign breast diseases, and physical activity. The increment is 50 g of increase in
carbohydrate intake for the HRs and 95% CIs.
2
For the tests of interaction.

fat and protein) while holding constant the intakes of total energy.
Thus, we were able to compare the effect of dietary carbohydrate
with that of dietary fat and protein and found that dietary carbohydrate intake was responsible for the increase in breast cancer
risk in premenopausal women.
The proposed mechanisms for a possible etiologic role of dietary
carbohydrate intake in breast cancer risk are related to the development or exacerbation of insulin resistance or chronic
hyperinsulinemia (1, 3, 37). Insulin increases cell proliferation,
and insulin receptors are expressed in normal and malignant breast
tissue; thus, insulin may play an important role in breast cancer
etiology (38, 39). In addition, insulin inhibits the synthesis of insulin-like growth factor binding protein I, increases the bioavailability of insulin-like growth factor I (IGF-I) (2, 40), and thus
increases the risk of breast cancer, particularly in premenopausal
women (2, 4143).
Previous prospective studies have generally shown no association between dietary carbohydrate intake and breast cancer risk
(1217). A recent prospective study (44), however, reported that
starch-rich foods were associated with an increased risk of breast
cancer and ovarian cancer (odds ratio: 1.85; 95% CI: 1.37, 2.48)
for the highest consumption quartile compared with the lowest
quartile. Another prospective study (12) reported that body weight
modified the association of carbohydrate intake with breast cancer

risk and that carbohydrate intake was positively related to breast

cancer risk only among overweight premenopausal women. The
authors of the study postulated that obesity, as an important determinant of insulin resistance, might exaggerate adverse metabolic responses related to carbohydrate intake. In our study, we did
not find that body weight or other risk factors for breast cancer
modified the association between carbohydrate intake and breast
cancer risk. However, we found that the effect of carbohydrates on
breast cancer risk was modified by menopausal status and age,
with the increased risk associated with high dietary carbohydrate
intakes being observed only among premenopausal women or
women younger than 50 y. Previous studies have shown that
plasma IGF-I is particularly relevant to breast cancer risk in premenopausal women (4143), given that estradiol enhances the
action of IGF-I in the breast (45). Our findings support the hypothesis that the effect of carbohydrate intake on breast cancer risk
may be mediated by insulin and IGF-I.
Traditional ecological analysis (25) and the fact that while
carbohydrate intake has been decreasing in Shanghai, breast cancer
incidence has been increasing (26, 27) indicate a negative association between carbohydrate intake and breast cancer. In contrast,
however, we found a positive association between carbohydrate
intake and breast cancer in premenopausal women and women
younger than 50 y in this study. The heterogeneity of the association

288

WEN ET AL

between carbohydrate intake and breast cancer across age groups

(effect modification by age group) may be the reason for such a
large ecological bias (46).
In this study, glycemic load, a product of carbohydrate quantity
and quality, was highly correlated with total carbohydrate intake
(correlation coefficient 0.95). Therefore, glycemic load has an
effect on breast cancer risk similar to that of total carbohydrate
intake. Glycemic index, on the other hand, was not found to be
associated with breast cancer risk. A recent study (47) reported
similar findings in which endometrial cancer risk was associated
with total carbohydrate intake and glycemic load, but not with
glycemic index. A possible explanation for these findings was
ascribed to the narrow range and centered distribution of the observed glycemic index values present in the data (47, 48). In our
data, the glycemic index range was also narrow (interquartile
range 6.6; CV 7.4%), and 2 food groups, rice and noodles
(73.9%) and steamed bread (73%), contributed to .80% of the
dietary glycemic load in the study population (7). In addition, the
well-known carbohydrates (eg, sugar, fructose, or fat-containing
foods) have a low or medium glycemic index. It is difficult to
determine the effect of glycemic index with this type of distribution.
The strengths of this study include its large sample size, its
prospective cohort design, its extraordinarily high participation
rate (92.7%) at baseline recruitment, and its collection of detailed
information on many potential confounders. As with any epidemiologic study using an FFQ, a potential limitation is that the
assessment of dietary intake is prone to measurement errors.
However, dietary intake was measured twice, first during the
baseline interview and then at the first follow-up survey. We analyzed the average dietary intakes from the 2 measurements and
found similar results when only dietary intakes from the baseline
interview were analyzed. For example, the hazard ratios (and 95%
CIs) for every 50-g increase in carbohydrate intake in premenopausal women were 1.59 (1.20, 2.10) when analyzing the baseline
data and 1.51 (1.23, 1.86) when analyzing the averaged data (data
not shown in the tables). This comparison and the other strengths of
this study may alleviate some concerns about dietary measurement
errors. Another limitation of this study was that we did not analyze
the effect of specific types of carbohydrates (eg, fructose and sucrose), because we were not able to derive these variables from the
FFQ data. Some studies have suggested that a high intake of
fructose and sucrose may increase the risk of cancer (49, 50).
In conclusion, our data suggest no association of dietary fiber
with breast cancer risk and a positive linear association of dietary
carbohydrate intake and glycemic load with breast cancer risk in
premenopausal women or women younger than 50 y.
The authors responsibilities were as followsWW, XOS, GY, B-TJ, Y-TG,
and WZ: study design; HL and Y-TG: data acquisition; WW, XOS, B-TJ, HC,
and WZ: data analysis and interpretation; WW: draft of manuscript; and XOS,
HC, B-TJ, and WZ: critical revision of manuscript. None of the authors had
any personal or financial conflicts of interest.

REFERENCES
1. Kaaks R, Lukanova A. Energy balance and cancer: the role of insulin
and insulin-like growth factor-I. Proc Nutr Soc 2001;60:91106.
2. Kaaks R. Nutrition, insulin, IGF-1 metabolism and cancer risk: a summary
of epidemiological evidence. Novartis Found Symp 2004;262:24760.
3. Gupta K, Krishnaswamy G, Karnad A, Peiris AN. Insulin: a novel factor
in carcinogenesis. Am J Med Sci 2002;323:1405.
4. Del Giudice ME, Fantus IG, Ezzat S, Keown-Eyssen G, Page D,
Goodwin PJ. Insulin and related factors in premenopausal breast cancer
risk. Breast Cancer Res Treat 1998;47:11120.

5. Holt SH, Miller JC, Petocz P. An insulin index of foods: the insulin
demand generated by 1000-kJ portions of common foods. Am J Clin
Nutr 1997;66:126476.
6. Pereira MA, Jacobs DR Jr, Pins JJ, et al. Effect of whole grains on insulin sensitivity in overweight hyperinsulinemic adults. Am J Clin Nutr
2002;75:84855.
7. Villegas R, Liu S, Gao YT, et al. Prospective study of dietary carbohydrates, glycemic index, glycemic load, and incidence of type 2 diabetes mellitus in middle-aged Chinese women. Arch Intern Med 2007;
167:23106.
8. Augustin LS, Dal ML, La VC, et al. Dietary glycemic index and glycemic load, and breast cancer risk: a case-control study. Ann Oncol
2001;12:15338.
9. Potischman N, Coates RJ, Swanson CA, et al. Increased risk of early-stage
breast cancer related to consumption of sweet foods among women less
than age 45 in the United States. Cancer Causes Control 2002;13:93746.
10. Tavani A, Giordano L, Gallus S, et al. Consumption of sweet foods and
breast cancer risk in Italy. Ann Oncol 2006;17:3415.
11. Silvera SA, Jain M, Howe GR, Miller AB, Rohan TE. Dietary carbohydrates and breast cancer risk: a prospective study of the roles of
overall glycemic index and glycemic load. Int J Cancer 2005;114:6538.
12. Cho E, Spiegelman D, Hunter DJ, Chen WY, Colditz GA, Willett WC.
Premenopausal dietary carbohydrate, glycemic index, glycemic load,
and fiber in relation to risk of breast cancer. Cancer Epidemiol Biomarkers Prev 2003;12:11538.
13. Higginbotham S, Zhang ZF, Lee IM, Cook NR, Buring JE, Liu S. Dietary glycemic load and breast cancer risk in the Womens Health Study.
Cancer Epidemiol Biomarkers Prev 2004;13:6570.
14. Holmes MD, Liu S, Hankinson SE, Colditz GA, Hunter DJ, Willett WC.
Dietary carbohydrates, fiber, and breast cancer risk. Am J Epidemiol
2004;159:7329.
15. Horn-Ross PL, Hoggatt KJ, West DW, et al. Recent diet and breast
cancer risk: the California Teachers Study (USA). Cancer Causes Control
2002;13:40715.
16. Jonas CR, McCullough ML, Teras LR, Walker-Thurmond KA, Thun
MJ, Calle EE. Dietary glycemic index, glycemic load, and risk of incident breast cancer in postmenopausal women. Cancer Epidemiol
Biomarkers Prev 2003;12:5737.
17. Nielsen TG, Olsen A, Christensen J, Overvad K, Tjonneland A. Dietary
carbohydrate intake is not associated with the breast cancer incidence
rate ratio in postmenopausal Danish women. J Nutr 2005;135:1248.
18. Cohen LA. Dietary fiber and breast cancer. Anticancer Res 1999;
19:36858.
19. Gerber M. Fibre and breast cancer. Eur J Cancer Prev 1998;7(suppl
2):S637.
20. Challier B, Perarnau JM, Viel JF. Garlic, onion and cereal fibre as
protective factors for breast cancer: a French case-control study. Eur J
Epidemiol 1998;14:73747.
21. Ronco A, De SE, Boffetta P, Deneo-Pellegrini H, Mendilaharsu M,
Leborgne F. Vegetables, fruits, and related nutrients and risk of breast
cancer: a case-control study in Uruguay. Nutr Cancer 1999;35:1119.
22. Cade JE, Burley VJ, Greenwood DC. Dietary fibre and risk of breast
cancer in the UK Womens Cohort Study. Int J Epidemiol 2007;
36:4318.
23. Suzuki R, Rylander-Rudqvist T, Ye W, Saji S, Adlercreutz H, Wolk A.
Dietary fiber intake and risk of postmenopausal breast cancer defined by
estrogen and progesterone receptor status: a prospective cohort study
among Swedish women. Int J Cancer 2008;122:40312.
24. Terry P, Jain M, Miller AB, Howe GR, Rohan TE. No association among
total dietary fiber, fiber fractions, and risk of breast cancer. Cancer
Epidemiol Biomarkers Prev 2002;11:15078.
25. Willett WC. Diet and breast cancer. J Intern Med 2001;249:395411.
26. Chen Z, Shu XO, Yang G, et al. Nutrient intake among Chinese women
living in Shanghai, China. Br J Nutr 2006;96:3939.
27. Jin F, Devesa SS, Chow WH, et al. Cancer incidence trends in urban
shanghai, 1972-1994: an update. Int J Cancer 1999;83:43540.
28. Shu XO, Yang G, Jin F, et al. Validity and reproducibility of the food
frequency questionnaire used in the Shanghai Womens Health Study.
Eur J Clin Nutr 2004;58:1723.
29. Zheng W, Chow WH, Yang G, et al. The Shanghai Womens Health
Study: rationale, study design, and baseline characteristics. Am J Epidemiol 2005;162:112331.
30. Lohman TG, Roche AF, Martorell R. Anthropometric standardization
reference manual. Champaign, IL: Human Kinetics Books.

CARBOHYDRATES, FIBER, AND BREAST CANCER RISK

31. Wang GY, Shen ZP. Chinese food composition tables. Beijing, Peoples
Republic of China: Peoples Health Publishing House, 1991.
32. Foster-Powell K, Holt SH, Brand-Miller JC. International table of
glycemic index and glycemic load values: 2002. Am J Clin Nutr 2002;
76:556.
33. Wolever TM, Jenkins DJ, Jenkins AL, Josse RG. The glycemic index:
methodology and clinical implications. Am J Clin Nutr 1991;54:84654.
34. Hosmer DW, Lemeshow S. Applied survival analysis: regression modeling of time to event data. New York, NY: John Wiley & Sons, 1999.
35. Korn EL, Graubard BI, Midthune D. Time-to-event analysis of longitudinal follow-up of a survey: choice of the time-scale. Am J Epidemiol
1997;145:7280.
36. Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake in
epidemiologic studies. Am J Clin Nutr 1997;65:1220S8S.
37. Ludwig DS. The glycemic index: physiological mechanisms relating to
obesity, diabetes, and cardiovascular disease. JAMA 2002;287:241423.
38. Milazzo G, Giorgino F, Damante G, et al. Insulin receptor expression
and function in human breast cancer cell lines. Cancer Res 1992;52:
392430.
39. Papa V, Milazzo G, Goldfine ID, Waldman FM, Vigneri R. Sporadic
amplification of the insulin receptor gene in human breast cancer. J
Endocrinol Invest 1997;20:5316.
40. Conover CA, Lee PD, Kanaley JA, Clarkson JT, Jensen MD. Insulin
regulation of insulin-like growth factor binding protein-1 in obese and
nonobese humans. J Clin Endocrinol Metab 1992;74:135560.
41. Hankinson SE, Willett WC, Colditz GA, et al. Circulating concentrations of insulin-like growth factor-I and risk of breast cancer. Lancet
1998;351:13936.

289

42. Toniolo P, Bruning PF, Akhmedkhanov A, et al. Serum insulin-like

growth factor-I and breast cancer. Int J Cancer 2000;88:82832.
43. Yu H, Jin F, Shu XO, et al. Insulin-like growth factors and breast cancer
risk in Chinese women. Cancer Epidemiol Biomarkers Prev 2002;11:
70512.
44. Edefonti V, Decarli A, La VC, et al. Nutrient dietary patterns and the
risk of breast and ovarian cancers. Int J Cancer 2008;122:60913.
45. Ruan W, Catanese V, Wieczorek R, Feldman M, Kleinberg DL. Estradiol
enhances the stimulatory effect of insulin-like growth factor-I (IGF-I)
on mammary development and growth hormone-induced IGF-I messenger ribonucleic acid. Endocrinology 1995;136:1296302.
46. Rothman KJ, Greenland S. Modern epidemiology. Philadelphia, PA:
Lippincott-Raven Publishers, 1998.
47. Cust AE, Slimani N, Kaaks R, et al. Dietary carbohydrates, glycemic
index, glycemic load, and endometrial cancer risk within the European
Prospective Investigation into Cancer and Nutrition Cohort. Am J Epidemiol 2007;166:91223.
48. Flood A, Peters U, Jenkins DJ, et al. Carbohydrate, glycemic index, and
glycemic load and colorectal adenomas in the Prostate, Lung, Colorectal,
and Ovarian Screening Study. Am J Clin Nutr 2006;84:118492.
49. Higginbotham S, Zhang ZF, Lee IM, et al. Dietary glycemic load and
risk of colorectal cancer in the Womens Health Study. J Natl Cancer
Inst 2004;96:22933.
50. Nothlings U, Murphy SP, Wilkens LR, Henderson BE, Kolonel LN.
Dietary glycemic load, added sugars, and carbohydrates as risk factors
for pancreatic cancer: the Multiethnic Cohort Study. Am J Clin Nutr
2007;86:1495501.